Comparison of Estimation Procedures for First Order Moving Average Multiple Time Series.
STATE UNIV OF NEW YORK AT BUFFALO AMHERST STATISTICAL SCIENCE DIV
Pagination or Media Count:
This dissertation provides an empirical examination of four estimation procedures associated with the estimation of the matrix parameters for a vector moving average process. An important role in this examination is the development of the relationships between the parametric representation for both the vector autoregressive and moving average schemes. This leads to the examination, comparison and utilization of autoregressive order determination and fitting criteria. Spectral plotting and non-linear numerical techniques are also employed. A comparison of the estimation procedures is provided based upon both theoretical and practical criterion.
- Statistics and Probability
- Computer Programming and Software